An SEO audit with 800 findings looks thorough and often changes surprisingly little. The number does not tell you which pages matter for the business or which problems actually cost visibility. A good audit therefore answers three questions: What is demonstrably wrong? Which important pages and queries are affected? And which specific action is justified?
After reading, you can build an audit that ends in a few evidenced decisions: you check technical health, content, backlinks and visibility separately, prioritize by impact and give every task an acceptance criterion. It ends with a compact checklist.
Preparation: start with a question
"Let's do an SEO audit" is not a brief. A good question narrows the scope and determines which data you need. Examples:
- Why have category pages been losing impressions for three months?
- Is the site technically clean after the relaunch?
- Which guides bring no visitors, and why?
- Are the new language versions set up correctly?
Before you start, also record exactly what is being examined: domains and subdomains, included and excluded directories, the comparison period in Search Console and known changes such as releases or migrations. Without these details, a finding is hard to interpret later.
The tools you need
A solid audit needs only a few tools, used deliberately:
- Google Search Console for indexing, queries, clicks and impressions of your own site.
- A crawler that captures every reachable URL with status code, metadata, canonical and internal links.
- Web analytics such as Google Analytics, to see which pages bring leads or revenue.
- Ranking and keyword data to assess demand and competition.
- A spreadsheet or ticket system where every finding lands with a status and an owner.
More important than the number of tools is that all data belongs to the same period and the same scope.
Layer 1: Technical health
The foundation is a complete crawl of the site. It shows what search engines find when they visit. Check in particular:
- Status codes. Error pages with 4xx and 5xx, especially when they are linked internally.
- Redirects. Chains, loops and internal links that only reach their target via a redirect.
- Indexability. Pages with noindex or robots.txt blocks that are supposed to rank.
- Canonicals. Missing annotations and canonicals pointing to other or unreachable pages. The glossary covers the basics of the canonical tag.
- Internal linking. Important pages with few internal links or a high click depth.
- Sitemap. URLs in the sitemap that redirect, return errors or carry noindex.
- Metadata. Missing, duplicate, too long or too short titles and descriptions.
- Structured data. Broken or incomplete markup on page types meant to produce rich results. More on structured data.
- Performance. Core Web Vitals from real user data, complemented by response times from the crawl.
- Rendering. Whether main content and links are in the HTML or only created by JavaScript.
Do not evaluate every finding on its own. Look for patterns: when all product pages share the same problem, there is almost always a template behind it, and a single fix resolves hundreds of findings.
In addition, the Page indexing report in Search Console shows which pages Google actually indexes. Our guide to Crawled - currently not indexed explains how to handle pages Google crawls but does not index.
Layer 2: Content
Technically clean pages do not rank automatically. The second layer checks whether the content delivers what searchers expect.
- Thin content. Pages with little text of their own, such as categories without an introduction or generated pages. More on thin content.
- Overlaps. Several pages answering the same question and competing with each other. The glossary explains how to spot keyword cannibalization.
- Outdated content. Guides with obsolete figures, screenshots or recommendations.
- Fit with search intent. If the results page shows how-to guides while your page sells a product, ranking will be hard.
- Trust signals. Author information, sources, first-hand experience and freshness, especially for topics where mistakes are costly. The background is covered under E-E-A-T.
No tool can do this layer for you. Take a sample per page type and assess it honestly. A structured content audit with fixed criteria helps make the assessment comparable across many pages.
Layer 3: Backlinks and off-page
Links from other websites remain an important signal. In an audit, this is less about the raw number than about changes and anomalies.
- Trend in referring domains. A clear decline over months can explain why rankings slip despite clean technical health.
- Lost links to important pages. If external links point to URLs that now return 404, you are throwing away their value. A matching redirect gets it back.
- Most-linked pages. They show which content others consider worth citing and are good starting points for internal links to important pages.
- Anchor texts. A natural link profile contains mostly brand and URL anchors. Many exact-match keyword anchors can point to bought links.
- Putting spam links in perspective. Google ignores most low-quality links on its own. According to Google, a disavow file mainly makes sense when there is, or is likely to be, a manual action for unnatural links.
The domain overview in Crawl Foundry shows backlinks, referring domains, anchor texts, the most-linked pages and the monthly trend. The documentation on backlinks and growth explains what the metrics mean.
Layer 4: Visibility and demand
The fourth layer connects the findings with real search data. This is where you see which problems actually cost visitors.
- Pages losing traffic. Compare clicks and impressions per page in Search Console with an earlier period.
- High impressions, few clicks. These pages rank but rarely get clicked. The title or description is often the reason.
- Missing topics. Important queries without a matching page.
- Rankings. How positions for your most important keywords develop. Our article on the rank tracker shows how to measure them reliably.
Read this data together with the technical layer. A page that is technically clean and still losing impressions has a content or competition problem. A page with a serious technical finding that hardly anyone searches for can wait.
Prioritizing findings
A plain severity list rarely leads to the right order. Four work classes are more useful:
- Fix now. A clear error with a large impact and a safe fix, such as an important page set to noindex.
- Investigate. A relevant signal whose cause is still open, such as falling impressions without a technical finding.
- Review editorially. The technology works, but the content or role of the page is questionable.
- Monitor or accept. An intended state or a change with little benefit.
Within a class, two things determine the order: how important the affected pages are and how solid the evidence is. A template error on every product page beats a warning on three archive pages. Our article on technical SEO issues in Site Audit goes deeper into ordering crawl findings.
An example: auditing a drop in impressions
An online retailer for garden furniture sees in Search Console that its category pages have been getting fewer impressions for three months. The audit question is therefore: why are the category pages losing visibility, and what can be done about it?
At first glance the crawl shows many findings, including missing alt text, long titles and a few broken links in the blog. Only two patterns matter for the question, though. First, the filtered versions of the category pages carry self-referencing canonicals and are listed in the sitemap, so hundreds of near-identical pages compete with the actual categories. Second, a template update three months ago moved the category introduction into a tab that only loads with JavaScript.
The tasks almost write themselves: point the canonicals of the filter pages to the main category, remove the filter pages from the sitemap and put the introduction back into the HTML. The remaining findings go into the "monitor" class and are handled later. After implementation, a second crawl confirms the fixes, and impressions are monitored over the following weeks.
Writing tasks and verifying fixes
"Improve content" is not an actionable task. A good task names the affected URLs or page group, the finding, the suspected cause, the desired change, the owner and an acceptance criterion. The acceptance criterion is the most important part, because it defines how you will know later that the problem is solved.
After implementation, run a second crawl with the same scope. Only the comparison shows whether findings have really disappeared or whether new ones have appeared. Rankings and Search Console data take longer to respond. Schedule a check after a few weeks.
Checklist
- Audit question and scope defined
- Complete crawl with a saved snapshot
- Status codes, redirects and indexability checked
- Canonicals, sitemap and internal linking checked
- Metadata and structured data checked
- Core Web Vitals from real user data reviewed
- Sample per page type assessed for content
- Overlaps and outdated content flagged
- Referring domains, lost links and anchor texts reviewed
- Search Console data evaluated per page
- Findings sorted into work classes
- Tasks written with acceptance criteria
- Verification crawl scheduled
How Crawl Foundry helps with an SEO audit
Crawl Foundry Site Audit handles the technical part. It crawls the site and summarizes the results in a health score weighted by category: crawlability and availability at 25 percent, indexability and canonicals at 20, content and metadata as well as links and architecture at 15 each, structured and international data as well as performance at 10 each, and security at 5 percent. Every finding traces back to specific URLs, and crawl comparison shows after implementation what is new, fixed or regressed. The documentation on overview and statistics explains how the score is calculated and when two crawls are comparable.
From findings to a prioritized task list
An SEO audit is valuable when it leads to a few clear decisions. Start with a question, check technical health, content, backlinks and visibility separately and then read them together. Prioritize by impact and evidence, write tasks with acceptance criteria and verify the result.
As a first step, write down the one business question your audit should answer.